TWI827974B - Virtual function performance analyzing system and analyzing method thereof - Google Patents

Abstract

A virtual function performance analyzing system and an analyzing method thereof are disclosed. The virtual function performance analyzing method includes the following. Monitoring at least one virtual function of a virtual network function application in a virtual platform, the virtual platform includes at least one physical resource and/or at least one virtual resource. Monitoring and recording in the virtual platform, an actual value of each performance indicator of each physical resource and/or each virtual resource、and each virtual function. Comparing the actual value of each performance indicator with the corresponding expected value and/or threshold value to obtain a comparison result. Analyzing system performance according to the comparison result.

Description

Virtual function performance analysis system and analysis method

This case is about a virtual function performance analysis system and its analysis method, especially an analysis system and its analysis method that analyzes the execution performance of virtual function functions under the architecture of a virtual platform.

With the evolution of network communication architecture and software technology, various physical computing resources or storage resources can be operated in the virtual platform architecture through virtualization. Through the virtualization architecture, the user's terminal program (virtual network function application) can flexibly access virtual computing resources or storage resources in the virtual platform architecture.

However, in the environment of the virtual platform, the execution of virtual network function applications and the access and use of virtual resources are far more complicated than the physical environment. Furthermore, the virtual network function application may be freely available open source software or a program suite provided by a third party. If the virtual network function application operates abnormally, resulting in a decrease in the overall performance of the virtual platform, it is difficult for the user to identify the specific function within the virtual network function application that operates abnormally.

Therefore, technicians in related industries in this technical field are committed to analyzing the performance performance of functions (virtual function functions) within virtual network function applications, hoping to improve Accurately analyze the reasons that may cause the performance of the virtual platform to decrease, and identify the problem points that may cause the performance bottleneck of the virtual platform.

One embodiment of the present invention provides a virtual function performance analysis system, including a monitoring unit, a performance analysis unit, and a resource adjustment unit. The monitoring unit is used to monitor the performance of at least one virtual function function of the virtual network function application on the virtual platform. The virtual platform has at least one physical resource and at least one virtual resource. Each physical resource, each virtual resource and each Each of the virtual function functions has at least one performance index, and each performance index has an expected value and/or a threshold value. The monitoring unit monitors and records the actual value of each performance index. The performance analysis unit is used to compare the actual value of each performance indicator with its corresponding expected value and/or threshold to obtain the comparison results of each performance indicator and analyze system performance accordingly.

Another embodiment of the present application provides a virtual function performance analysis method, including the following steps. Monitoring at least one virtual function function of an application of a virtual network function on a virtual platform, the virtual platform having at least one physical resource and at least one virtual resource, each of the physical resources, each of the virtual resources and each of the virtual function functions Each of the formulas has at least one performance index, and each performance index has an expected value and/or a threshold value. The actual values of each performance indicator are monitored and recorded on the virtual platform. Compare the actual value of each performance indicator with its corresponding expected value and/or threshold to obtain the comparison result and analyze the system performance accordingly.

By reading the following drawings, detailed descriptions of various embodiments, and patent claims, other aspects and advantages of this case can be seen.

100A (or 100B): Virtual function performance analysis system

102:Test unit

104:Monitoring unit

106:Performance analysis unit

108: Resource adjustment unit

200,230,260: Virtual network function application

202,204,206,302:Virtual function function

t1: initial time point

t2: end time point

202a,204a:queuing

T_02, T_04, T_06: Execution time

300:Test platform

400:Virtual platform

402:Entity CPU

404:Physical memory

406:Physical network bandwidth

408:Physical disk input/output

410:Virtual layer

412:Virtual CPU

412a, 412b, 412c: virtual resources

414:Virtual memory

416:Virtual network bandwidth

418:Virtual disk input/output

600, 600B, 700, 800, 900: Example of virtual function performance analysis method

602~628: Steps

Figure 1A is a block diagram of an embodiment of the virtual function performance analysis system disclosed in this case.

Figure 1B is a block diagram of another embodiment of the virtual function performance analysis system disclosed in this case.

Figure 2 is a schematic diagram of an embodiment of a performance indicator of a virtual function function disclosed in this case.

Figure 3A is a schematic diagram of an operational embodiment of the test unit disclosed in this case.

Figure 3B is a schematic diagram of an embodiment of a performance indicator of a virtual function function disclosed in this case.

Figure 4A is a schematic diagram of an embodiment of the physical resources and virtual resources of the virtual platform disclosed in this case.

Figure 4B is a schematic diagram of an embodiment of virtual resource configuration of the virtual platform disclosed in this case.

Figure 5 is a schematic diagram of an operational embodiment of the monitoring unit disclosed in this case.

Figures 6A to 6C are flow charts of one embodiment of the virtual function performance analysis method disclosed in this case.

Figure 7 is a flow chart of another embodiment of the virtual function performance analysis method disclosed in this case.

Figure 8 is a flow chart of another embodiment of the virtual function performance analysis method disclosed in this case.

Figure 9 is a flow chart of another embodiment of the virtual function performance analysis method disclosed in this case.

The technical terms in this specification refer to the idioms in the technical field. If there are explanations or definitions for some terms in this specification, the explanation or definition of this part of the terms shall prevail. Each embodiment of the present disclosure has one or more technical features. To the extent possible, a person with ordinary skill in the art can selectively implement some or all of the technical features in any embodiment, or selectively combine some or all of the technical features in these embodiments.

Figure 1A is a block diagram of an embodiment 100A of a virtual function performance analysis system disclosed in this case. Referring to Figure 1A, the virtual function performance analysis system 100A includes a monitoring unit 104, a performance analysis unit 106, and a resource adjustment unit 108. The virtual function performance analysis system 100A performs performance analysis on the application (application) of the virtual network function (Virtual Network Function, VNF) and the performance of each virtual function function executed on the network function virtualization infrastructure (NFVI). Analysis and monitoring.

In the virtual platform of the actual field environment where the virtual network function application is actually executed, the monitoring unit 104 can monitor the performance performance of at least one virtual function function inside the virtual network function application. More specifically, the virtual network function application and each virtual function function have one or more performance indicators, each performance indicator The standard reflects the performance of virtual network function applications and each virtual function function executed on the virtual platform. Each performance indicator has an expected value, and the expected value can be a system default value or defined by the user. In addition, machine learning can also be used to observe and analyze the long-term running status of each virtual function function on the virtual platform, so as to establish the expected value of each performance indicator. Moreover, the usage status of the virtual network function application's physical resources and virtual resources of the virtual platform are also performance indicators analyzed by the virtual function performance analysis system 100A.

The monitoring unit 104 monitors the execution of these virtual function functions, records the actual performance performance of each virtual function function in the virtual platform, and then collects and obtains the actual values of each performance indicator of each virtual function function. In addition, the monitoring unit 104 can also monitor the usage status of each virtual resource and each physical resource in the virtual platform by each virtual network function application, and thereby record the actual values of each performance indicator of each virtual resource and each physical resource.

The performance analysis unit 106 can analyze and compare the actual values of each performance indicator of each virtual resource, each physical resource and each virtual function function with the corresponding expected value, and analyze and compare the performance of each virtual network function application program on each virtual resource and each virtual function function. The usage status of each physical resource is compared to obtain a comparison result, and the system performance is analyzed based on the comparison result to determine whether the virtual network function application and each virtual function function are operating abnormally, resulting in a decrease in the overall performance of the virtual platform. The performance analysis unit 106 can also pinpoint specific virtual function functions, virtual resources or physical resources that cause performance degradation to point out possible abnormal causes of system performance bottlenecks.

According to the pointed out abnormal cause, the resource adjustment unit 108 can be used to Adjust system resources on the execution status of the virtual network function application or each virtual function function, for example: expand the services of the virtual network function application to reduce the workload of the virtual function function that causes performance bottlenecks. Alternatively, the resource adjustment unit 108 may adjust the configuration of the physical resources or virtual resources of the virtual platform, for example, by expanding the physical resources or adjusting the configuration of the virtual resources, so as to allocate sufficient physical resources or virtual resources to the performance bottlenecks. The virtual network function application that the virtual function function belongs to.

Figure 1B is a block diagram of another embodiment 100B of the virtual function performance analysis system disclosed in this case. Please refer to Figure 1B. Compared with the virtual function performance analysis system 100A in Figure 1A, the virtual function performance analysis system 100B of this embodiment may further include a test unit 102. The testing unit 102 can test each virtual function function in the virtual network function application in advance on the test platform to evaluate the expected performance of each virtual function function (i.e., when physical or virtual resources are sufficient, Each virtual function function meets the expected normal performance performance), and the expected value of each performance indicator of each virtual function function is established accordingly.

In other words, unlike the virtual function performance analysis system 100A, the expected value of each performance index can be preset by the system or defined by the user. In the virtual function performance analysis system embodiment 100B, the expected value of each performance index is determined by the test unit. 102 Pre-built in the test platform or via machine learning.

The above is an overview of the virtual function performance analysis system 100A or 100B disclosed in this case in conjunction with Figures 1A and 1B. The following is an overview of each unit of the virtual function performance analysis system 100A or 100B in conjunction with Figures 2, 3A, 3B, 4 and 5. the specific reality Implementation methods are explained in detail.

Figure 2 is a schematic diagram of an embodiment of a performance indicator of a virtual function function disclosed in this case. Please refer to Figure 2. The virtual network function application 200 is the main monitoring and analysis target of the virtual function performance analysis system 100A (or 100B). The virtual network function application 200 can be called the "target virtual network function application". The virtual network function application 200 is, for example, a program (process) or virtual machine (virtual machine) executed in a virtual platform. The virtual network function application 200 may include a plurality of virtual function functions, for example, at least include a virtual function function 202, a virtual function function 204, and a virtual function function 206. Each of the virtual function functions 202, 204, and 206 may have one or more performance indicators, and these performance indicators may reflect the performance performance of the virtual function functions 202, 204, and 206 at each execution level in the virtual platform. The performance indicators of the virtual function functions 202, 204, and 206 may include, for example: "execution time", "execution frequency", "error occurrence rate" and "queue" of the virtual function functions 202, 204, and 206. column length" and so on, but is not limited to this.

First, take the performance index "execution time" of virtual function functions 202, 204, and 206 as an example. The execution time directly reflects the performance of the virtual function functions 202, 204, and 206. If the execution time is too long, the following situations may occur: the execution of the virtual function functions 202, 204, and 206 is blocked. The workload of 202, 204, 206 exceeds the load, there is competition in the use of physical resources of the virtual platform, or the physical resources and/or virtual resources of the virtual network function application 200 to which the virtual function functions 202, 204, 206 belong are configured. Insufficient resources.

More specifically, taking the virtual function function 202 as an example, the execution time T_02 of the virtual function function 202 is the time interval between the initial time point t1 and the end time point t2 of executing the virtual function function 202. The initial time point t1 is, for example, the time point when the virtual function function 202 is called, and the end time point t2 is, for example, the time point when the virtual function function 202 returns a result. In another example, the virtual function function 202 can be executed in a multi-thread environment, and the execution time T_02 of the virtual function function 202 is corresponding to each execution of the virtual function function 202 The execution time of each thread.

As mentioned above, in the virtual function performance analysis system 100B shown in Figure 1B, the expected values of each performance index of the virtual function function can be established in advance by the test unit 102, and Figure 3A shows the test unit disclosed according to this case Schematic diagram of an operational embodiment of 102. Referring to Figure 3A, the test unit 102 tests the virtual function functions 202, 204, and 206 of the virtual network function application 200 in the test platform 300 in advance. The test platform 300 is a platform with a clean room environment, such as a laboratory environment. The clean room environment of the test platform 300 is a simplified environment that minimizes environmental variables or external factors that may affect the execution of the virtual function functions 202, 204, and 206. Therefore, the performance of the virtual function functions 202, 204, and 206 on the test platform 300 is generally in line with user expectations. In the test platform 300, the data values of each performance indicator of the virtual function functions 202, 204, and 206 collected by the test unit 102 can be used as the expected value of each performance indicator (or can be called a "baseline value"). For example, virtual function functions 202, 204, and 206 are testing The data values of the execution times T_02, T_04, and T_06 in the platform 300 can be used as expected values of the execution times of the virtual function functions 202, 204, and 206 respectively.

More specifically, taking the execution time T_02 of the virtual function function 202 as an example, the test unit 102 can observe the execution time T_02 through tools such as eBPF. Or, in another example, when writing the code of the virtual function function 202, the user can establish an observation point in the code of the virtual function function 202 to record the time when the virtual function function 202 is called. point (initial time point t1) and the time point of returning the result (end time point t2), and the execution time T_02 is calculated based on the history record (log) of the observation point of the program code.

Under different environmental factors, the execution times T_02, T_04, and T_06 of the virtual function functions 202, 204, and 206 may change. Therefore, the test unit 102 can make statistics on the execution times T_02, T_04, and T_06 of the virtual function functions 202, 204, and 206 under different environmental factors to obtain the minimum, average, and maximum values of the execution times T_02, T_04, and T06. . For example, as shown in Table 1, in the test platform 300, the expected execution times T_02, T_04, and T_06 of the virtual function functions 202, 204, and 206 created by the test unit 102 are respectively: execution of the virtual function function 202 The minimum value of time T_02 is 50us, the average value is 100us, and the maximum value is 200us. The minimum value of the execution time T_04 of the virtual function function 204 is 30us, the average value is 50us, and the maximum value is 150us. Moreover, the minimum value of the execution time T_06 of the virtual function function 206 is 100us, the average value is 130us, and the maximum value is 250us.

On the other hand, referring to Figure 3B, the program logic of the virtual function functions 202, 204, and 206 may have a queue form. For example, the input terminal (input mechanism) of the virtual function function 202 may have a queue 202a, and the virtual function function 202 may have a queue 202a. The output (output mechanism) of function 204 may also have queue 204a. The "queue length" of the queues 202a and 204a of the virtual function functions 202 and 204 is also another important performance indicator.

The virtual function performance analysis system 100A (or 100B) can set a threshold for the queue length of the queue 202a and the queue 204a. The threshold can be a system default value or can be defined by the user. For example, the user (the developer of the virtual function function) can count and evaluate the reasonable queue length of the queue 202a and the queue 204a based on the data of past usage experience, and set the reasonable queue length as the performance indicator. Queue length" threshold. As shown in Table 1, the queue length threshold of the queue 202a at the input port of the virtual function function 202 is set to 200, for example. If the queue length of queue 202a exceeds the threshold when the virtual function function 202 is actually executed, it means that the virtual function function 202 operates abnormally. On the other hand, for the queue at the output of virtual function function 204 For example 204a, if the queue length of queue 204a is too long, it means that another virtual function function 302 that uses queue 204a as an input may operate abnormally.

Moreover, the virtual function functions 202, 204, and 206 may also have other performance indicators, such as "execution frequency" and "error occurrence rate". Among them, the "execution frequency" of the virtual function functions 202, 204, and 206 is the number of execution times of the virtual function functions 202, 204, and 206 per unit time. In addition, if the virtual function functions 202, 204, and 206 are executed in a multi-thread environment, the above performance indicators "queue length", "execution frequency" and "error occurrence rate" are for the virtual function functions 202, 204 , 206 corresponds to the queue length, execution frequency and error occurrence rate of each execution thread.

In addition to predicting the expected values or thresholds of each performance indicator of the virtual function functions 202, 204, and 206 by testing and evaluating the test unit 102 in the test platform 300, defining them by the user or using system default values, they can also be predicted by Built by machine learning. For example, the actual running data values of each performance indicator of the virtual function functions 202, 204, and 206 can be recorded for a long time, and then the expected value or threshold value of each performance indicator can be analyzed and counted through machine learning methods.

In addition to the performance indicators such as "execution time", "queue length", "execution frequency" and "error rate" of the virtual function functions 202, 204, and 206, on the other hand, the virtual function functions 202, 204, The "usage status" of the physical resources or virtual resources of the virtual platform by the virtual network function application 200 to which 206 belongs can also reflect the performance performance of the virtual network function application 200. Therefore, the "usage status" of physical resources or virtual resources by the virtual network function application 200 is also a performance indicator to be analyzed by the virtual function performance analysis system 100A (or 100B). Figure 4A is a schematic diagram of an embodiment of physical resources and virtual resources of the virtual platform 400 disclosed in this case. Referring to Figure 4A, the virtual platform 400 is the field environment where the virtual network function application 200 is actually executed. The physical resources of the virtual platform 400 at least include: a physical central processing unit (CPU) 402, a physical memory (memory) 404, a physical network bandwidth (network bandwidth) 406, and a physical disk input/output (disk I/O). 408.

On the other hand, under the architecture of the virtual platform 400, the above physical resources can be virtualized through the virtual layer 410 and become corresponding virtual resources. The virtual resources corresponding to the physical CPU 402, the physical memory 404, the physical network bandwidth 406, and the physical disk input/output 408 are: virtual CPU 412, virtual memory 414, and virtual network bandwidth 416 , virtual disk input/output 418. Moreover, the virtual platform 400 can configure the above virtual resources to the virtual network function application 200 and other virtual network function applications. Please also refer to Figure 4B, which illustrates a schematic diagram of an embodiment of virtual resource configuration of the virtual platform 400 disclosed in this case. Taking the virtual CPU 412 in the virtual resource as an example, the virtual CPU 412 may include at least a virtual resource 412a, a virtual resource 412b, and a virtual resource 412c. Virtual resource 412a may be configured to virtual network function application 200, virtual resource 412b may be configured to virtual network function application 230, and virtual resource 412c may be configured to virtual network function application 260.

Please also refer to Table 2, which illustrates multiple performance indicators related to the "usage status" of physical resources and virtual resources of the virtual platform 400. Broadly speaking, the broad performance indicators of physical resources and virtual resources are physical resources and virtual resources. "usage status". More specifically, the performance indicator types related to the "usage status" of physical resources and virtual resources may include: "usage", "error events", "resource saturation", etc. of physical resources and virtual resources. Taking the performance indicator type "Usage" as an example, the performance indicator details of "Usage" of physical resources and virtual resources can include: the respective usage of a single CPU/memory/disk, and the average usage of the entire system , the traffic and bandwidth received or transmitted by the network interface, etc.

For the performance indicators listed in Table 2, the virtual function performance analysis system 100A (or 100B) can establish expected values and/or thresholds of these performance indicators based on system default values or user definitions. For example, as shown in Table 2-1: the virtual network function application 200 has a virtual configuration for the configured virtual CPU 412. The expected value of "usage" of resource 412a is 22%, and the expected value of "usage" of virtual resource 412b of virtual CPU 412 configured by virtual network function application 230 is 25%. The virtual network function The expected "usage" of the virtual resource 412c of the configured virtual CPU 412 by the application 260 is 15%. Moreover, the expected value of the total usage of the physical CPU 402 by the virtual network function applications 200, 230, and 260 is 62%.

Figure 5 is a schematic diagram of an operational embodiment of the monitoring unit 104 disclosed in this case. Referring to Figure 5, when the virtual function functions 202, 204, and 206 of the virtual network function application 200 are actually executed on the virtual platform 400, the monitoring unit 104 monitors and collects each performance indicator of the virtual function functions 202, 204, and 206. actual value. Taking the performance index "execution time" as an example, the actual values of the execution times T_02, T_04, and T_06 of the virtual function functions 202, 204, and 206 are shown in Table 3: The actual values of the execution time T_02 of the virtual function function 202 The minimum value is 55us, The average value is 103us and the maximum value is 205us. The actual value of the execution time T_04 of the virtual function function 204 has a minimum value of 27us, an average value of 49us, and a maximum value of 130us. The actual value of the execution time T_06 of the virtual function function 206 has a minimum value of 110us, an average value of 135us, and a maximum value of 270us. On the other hand, taking the performance index "queue length" as an example, the actual value of the queue length of the queue 202a of the virtual function function 202 monitored by the monitoring unit 104 is 250, and the queue length of the queue 204a of the virtual function function 204 is 250. The actual value of the column length is 250.

On the other hand, as mentioned above, the monitoring unit 104 also actually monitors and records the virtual network function application 200 (ie, the target virtual network function application) and other virtual network function applications 230 in the virtual platform 400 , 260 for the actual usage status of each virtual resource and each physical resource of the virtual platform 400 to obtain the actual value of each performance index of each physical resource and each virtual resource. Taking the performance indicator "usage" as an example, Table 4 shows the virtual network recorded by the monitoring unit 104 Actual values of the usage of the virtual CPU 412 and the physical CPU 402 by the road function applications 200, 230, 260.

In Table 4, the actual usage of the virtual CPU 412 (virtual resource 412a) configured or allocated by the virtual network function application 200 is 20%, and the virtual network function application 230 uses 20% of the configured or allocated virtual CPU 412 (virtual resource 412a). The actual usage of the virtual CPU 412 (virtual resource 412b) is 22%, and the actual usage of the virtual network function application 260 for the configured or assigned virtual CPU 412 (virtual resource 412c) is 18 %. In addition, the actual total usage of the physical CPU 402 by the virtual network function applications 200, 230, and 260 is 60%.

To sum up, the monitoring unit 104 can monitor and record each virtual function function inside the virtual network function application 200 (the target virtual network function application). The actual values of each performance indicator ("execution time", "queue length", "execution frequency", or "error occurrence rate", etc.) of equations 202, 204, and 206. Moreover, the monitoring unit 104 can also monitor and record the usage status of each physical resource and each virtual resource of the virtual platform 400 by the virtual network function application 200 and other virtual network function applications 230 and 260, so as to record the usage status of each physical resource and each virtual resource. The actual value of each performance indicator ("usage", "error event", or "resource saturation", etc.) of each virtual resource.

Accordingly, the performance analysis unit 106 analyzes the performance indicators of the virtual function functions 202, 204, and 206 of the virtual network function application 200 and the performance indicators of each physical resource and each virtual resource of the virtual platform 400, and compares each performance indicator. The actual value and expected value and/or threshold value of the performance indicator are used to determine whether the virtual function functions 202, 204, and 206 of the virtual network function application 200 are operating abnormally, or the usage status of each physical resource/each virtual resource is abnormal, And can further analyze the performance bottlenecks and causes that cause abnormalities. Then, the resource adjustment unit 108 can adjust the services of the virtual network function application 200 or the virtual function functions 202, 204, 206, adjust the configuration to the virtual network function application 200 and/or other virtual network function applications 230 , 260 virtual resources, or expand the physical resources of the virtual platform 400. For detailed operations of the performance analysis unit 106 and the resource adjustment unit 108, please refer to the various embodiments of the virtual function performance analysis method disclosed below.

Figures 6A to 6C are flow charts of an embodiment 600 of the virtual function performance analysis method disclosed in this case. Please refer to Figure 6A first: First, in step 601, according to the system default value definition or according to the user definition, for the main parameters to be monitored and analyzed, Set the expected values and/or thresholds of each performance indicator of the virtual function functions 202, 204, and 206 within the target virtual network function application 200 (i.e., the target virtual network function application), and set the virtual platform 400, each physical resource (physical CPU 402, physical memory 404, physical network bandwidth 406, physical disk input/output 408) and each virtual resource (virtual CPU 412, virtual memory 414, virtual The expected values and/or thresholds of various performance indicators (network bandwidth 416, virtual disk input/output 418). Among them, the performance indicators of the virtual function functions 202, 204, and 206 include, for example: "execution time", "queue length", "execution frequency" and "error occurrence rate", etc. The performance indicators of each physical resource and each virtual resource in the virtual platform 400 include, for example: "usage", "error events", "resource saturation", and so on.

Next, in step 606, the virtual function functions 202, 204, and 206 of the virtual network function application 200 are actually monitored in the actual field environment of the virtual platform 400. The virtual network function application 200 uses virtual resources and physical resources of the virtual platform 400 .

Next, in step 608, the monitoring unit 104 of the virtual function performance analysis system 100A (or 100B) monitors and records the actual values of each performance index of each virtual function function 202, 204, 206 in the virtual platform 400.

Next, in step 610, the monitoring unit 104 monitors and records the virtual network function application 200 and other virtual network function applications 230 and 260 in the virtual platform 400 for each physical resource and each virtual resource of the virtual platform 400. The actual usage status of simulated resources (such as usage) is used to establish the actual values of each performance indicator of each physical resource and each virtual resource. Then, step 613 in Figure 6C is executed.

In the method embodiment 600 of Figure 6A described above, the expected values of each performance indicator of each physical resource and each virtual resource, as well as the expected values of each performance indicator of the virtual function functions 202, 204, 206 and/ Or threshold, it is preset based on the system default value or user definition. However, in the method embodiment 600B shown in FIG. 6B , the expected values of each performance index of the virtual function functions 202 , 204 , and 206 can also be pre-established in the test platform 300 by the test unit 102 . As shown in Figure 6B, before step 606, steps 602 and 604 may be performed first. In step 602, the virtual function functions 202, 204, and 206 of the target virtual network function application 200 are executed on the test platform 300 in a clean room environment. Then, in step 604, the test unit 102 collects and obtains the data values of each performance indicator of the virtual function functions 202, 204, and 206 based on the execution results of the virtual function functions 202, 204, and 206 on the test platform 300, and based on this These data values are used to establish expected values for each performance indicator. As for steps 606 to 610 of the method embodiment 600B in Figure 6B, they are completely the same as the method embodiment 600 in Figure 6A, and will not be repeated here.

Next, referring to Figure 6C, in step 613, through the performance analysis unit 106 of the virtual function performance analysis system 100A (or 100B), the performance indicators (for example: "usage") of each physical resource and each virtual resource are Compare the actual values of each performance indicator (such as "execution time", "queue length", etc.) of the virtual function functions 202, 204, and 206 with the corresponding expected values and/or thresholds to obtain a Compare the results, and determine whether the operation of the virtual function functions 202, 204, and 206 is abnormal based on the above comparison results.

For example, in the next step 614, the comparison result between the actual value of the performance indicator and the expected value and/or the threshold is: the actual value of one of the performance indicators "execution time" of the virtual function functions 202, 204, and 206. The values were roughly in line with the expected values, and these actual values were not much greater than the expected values. See Table 3 at the same time. For example, the actual values (including the minimum value, average value, and maximum value) of the execution time of virtual function functions 202, 204, and 206 are generally consistent with the expected execution time values shown in Table 1, indicating that the virtual function The actual values of the execution time of functions 202, 204, and 206 on the virtual platform 400 are generally in line with the expected values. Therefore, it is determined that the workload of the virtual function functions 202, 204, and 206 has not exceeded the load. At this time, step 616 may be executed to further analyze whether the actual values of other performance indicators exceed expected values and/or thresholds.

In step 616, the performance analysis unit 106 may compare the actual value of another performance indicator "queue length" with a threshold. The comparison result obtained in step 616 is: the actual value of "queue length" is much greater than the threshold value (see Table 3): the actual value of the queue length of virtual function 202 is 250, which is much greater than the threshold value in Table 1 The queue length threshold shown is 200). According to the above comparison results, it can be confirmed that the operation of the virtual function function 202 is abnormal, and the bottleneck of the operation performance of the virtual platform 400 can be located in the virtual function function 202 whose actual value of the queue length is much larger than the threshold.

Next, in step 618, the performance analysis unit 106 confirms that the workload of the virtual function function 202 whose queue length is greater than the threshold is overloaded. Therefore, the resource adjustment unit 108 may choose to adopt the following solution to overcome the problem of the virtual function function 202. Abnormal operation: adjust the services of the virtual network function application 200, or adjust the configuration of physical resources or virtual resources of the virtual platform 400.

In this embodiment, the resource adjustment unit 108 adjusts the services of the virtual network function application 200. For example, it can expand and execute other virtual network function applications with the same functions to share the virtual network function application 200. Workload of function 202. .

Figure 7 is a flow chart of another embodiment 700 of the virtual function performance analysis method disclosed in this case. This method embodiment 700 continues steps 601 to 613 of embodiment 600, so Figure 7 omits steps 601 to 613.

Referring to Figure 7, in step 614 of this embodiment 700, the comparison result between the actual value of each performance indicator and the expected value and/or threshold value is: the actual value of one of the performance indicators "execution time" is much greater than the expected value. (As shown in Table 5: the average value of the actual value of the execution time T_04 of the virtual function function 204 is 238us and the maximum value of 2530us, which is much larger than the average value of 50us and the maximum value of the expected execution time T_04 shown in Table 1. Value 150us). Therefore, the performance analysis unit 106 determines that the virtual function 204 operates abnormally, and the virtual function 204 is located as a bottleneck that causes performance degradation.

The reason why the virtual function 204 becomes a bottleneck may be that the usage of physical resources or virtual resources has approached the upper limit of the configuration or may tend to be overloaded. For example: insufficient physical resources and/or virtual resources configured for the virtual network function application 200 to which the virtual function function 204 belongs may cause the usage to tend to be overloaded; or, for example, in the case of multiple threads, the virtual network Other functions inside the application 200 His virtual function 202 or 206 competes with virtual function 204 to use virtual resources; or other virtual network function applications 230, 260 and virtual network function application 200 (target virtual network function application) Competition for use of physical resources. At this time, step 620 can be performed to further clarify the reason.

In step 620, the performance analysis unit 106 may evaluate whether the usage of each physical resource and/or each virtual resource is approaching the upper limit of the configuration or may tend to be based on the actual values of the performance indicators recorded for each physical resource and each virtual resource. overload. The analysis results are shown in Table 6: the actual usage of the virtual CPU 412 by the virtual network function application 200 (that is, the target virtual network function application) to which the virtual function function 204 belongs is 54%, which is far more than The expected value of 20% shown in Table 2-1 indicates that the usage of the virtual CPU 412 by the virtual network function application 200 has approached the upper limit of the configuration amount of the virtual resource 412a configured for the virtual network function application 200, or Usage may tend to overload. Therefore, virtual functions can be clarified The possible reason why the function 204 becomes a performance bottleneck is that the virtual platform 400 configures insufficient resources (virtual resource 412a) of the virtual CPU 412 of the virtual network function application 200 to which the virtual function function 204 belongs.

Next, in step 622, the resource adjustment unit 108 adjusts the virtual resource configuration of the virtual network function application 200. For example, the configuration of the virtual CPU 412 can be adjusted to use more virtual resources 412a in the virtual CPU 412. Configured for virtual network function application 200. Alternatively, the resource adjustment unit 108 can expand the physical resources and/or virtual resources of the virtual platform 400. Alternatively, the resource adjustment unit 108 can also adjust the services of the virtual network function application 200, for example, additionally expand and execute other virtual network function applications with the same functions to share the workload of the virtual network function application 200.

Figure 8 is a flow chart of another embodiment 800 of the virtual function performance analysis method disclosed in this case. This method embodiment 800 continues the steps of the embodiment 600. Steps 601 to 613 and step 614 of the embodiment 700 are therefore omitted in Figure 8 .

Referring to Figure 8, in step 620 of the method embodiment 800, the performance analysis unit 106 analyzes that the usage of each virtual resource by the virtual network function application 200 falls within the normal range (for example, as shown in Table 7: Virtual Network The actual usage of the virtual CPU 412 by the road function application 200 is 25%, which is only slightly larger than the expected value of 22% shown in Table 2-1, but not much more than the expected value of 22%), so it is judged that the virtual network The usage of the virtual resources of the road function application 200 does not approach the upper limit of the configuration or tends to be overloaded.

Next, step 624 may be performed to further analyze whether the total usage of physical resources by all virtual network function applications 200, 230, and 260 of the virtual platform 400 tends to be overloaded.

In step 624, the analysis results recorded by the performance analysis unit 106 are as shown in Table 7: the actual value of the total usage of the physical CPU 402 is 95%, which is close to the upper limit of the configuration and is almost overloaded, so it is determined that the virtual network Other virtual network function applications 230 and 260 other than the functional application 200 may be competing to use the physical CPU 402, and the usage of the physical CPU 402 by other virtual network function applications 230 and 260 is excessive. High resulting in overloading. Table 7 shows that the virtual network function application 230 uses 52% of the virtual CPU resources (the usage of 52% is a relatively high value), indicating that the virtual network function application 230 is competing for the use of physical CPU resources. Server 402 resources.

Next, in step 626, the resource adjustment unit 108 adjusts the virtual resource configuration of the virtual network function application 200. For example, the configuration of the virtual CPU 412 can be adjusted to use more virtual resources 412a in the virtual CPU 412. Configure the virtual network function application 200 so that the virtual network function application 200 has sufficient virtual resource configuration. Alternatively, to adjust the physical CPU 402 whose usage approaches the upper limit of the configuration or tends to be overloaded, for example, the physical resources and/or virtual resources of the virtual platform 400 can be expanded to add more physical CPUs.

Figure 9 is a flow chart of another embodiment 900 of the virtual function performance analysis method disclosed in this case. The virtual function performance analysis method embodiment 900 continues steps 601 to 610 of embodiment 600, step 614 of embodiment 700, and Step 620 of the embodiment 800, therefore steps 601 to 620 are omitted in Figure 9 .

Referring to Figure 9, in step 624 of the method embodiment 900, if the analysis result of the performance analysis unit 106 is: the usage of each virtual resource by the virtual network function application 200 does not approach the upper limit of the configuration or tends to is overloaded, and the total usage of each entity resource is not close to the upper limit of the configuration or tends to be overloaded (for example, as shown in Table 4: the actual usage of the virtual CPU 412 by the virtual network function application 200 is 20%, of which 22% lower than the expected value shown in Table 2-1; and the actual total usage of the physical CPU 402 is 60%, which is lower than the expected value 62% shown in Table 2-1), it can be judged that there is no A situation occurs where other virtual network function applications 230 and 260 compete for the use of physical resources, or the virtual function functions 202, 204, and 206 within the virtual network function application 200 compete for the use of virtual resources. Therefore, the performance analysis unit 106 determines that the reason why the execution time of the virtual function function 204 seriously exceeds the expected value and becomes a performance bottleneck has nothing to do with the configuration of physical resources and/or virtual resources.

Next, in step 628, the performance analysis unit 106 determines that the reason why the virtual function function 204 has become a performance bottleneck may be: an error occurs in the program logic of the program code of the virtual function function 204 itself, for example: the program of the virtual function function 204 An error occurred in the lock mechanism in the code. Or, there is a packet processing abnormality. For example, the packet loss in the virtual function function 204 causes the retransmission mechanism of the software protocol to retransmit the packet, resulting in a long delay in the service related to the virtual function function 204.

In summary, according to the above-mentioned different embodiments, the technical solution disclosed in this case is to use the virtual function performance analysis system 100A (or 100B) to analyze Multiple performance indicators (such as "execution time", "execution frequency", "queue length") of the virtual function functions 202, 204, and 206 of the virtual network function application 200 (i.e., the target virtual network function application) and "error occurrence rate"), and multiple performance indicators (such as "usage", "error events" and "resource saturation") of each physical resource and each virtual resource of the virtual platform 400 are analyzed. Moreover, each performance index of the virtual function functions 202, 204, and 206 of the virtual network function application 200 and the actual values of each performance index of each physical resource and virtual resource can be actually monitored on the virtual platform 400 in the actual field environment. . Then, analyze and compare whether the actual value of each of the above performance indicators is much greater than the expected value and/or threshold. If the actual value of each of the above performance indicators is much greater than the corresponding expected value and/or threshold, it means that this virtual function function ( For example: the workload of the virtual function function 202) is overloaded, or the physical resources and/or virtual resources allocated to the virtual network function application 200 to which this virtual function function belongs are insufficient, or other virtual network function applications The programs 230 and 260 compete to use physical resources, or the virtual function functions 202, 204, and 206 in the virtual network function application 200 compete to use virtual resources. The virtual function performance analysis system 100A (or 100B) automatically locates this "virtual function function", or the "physical resource" or "virtual resource" whose usage approaches the upper limit of the configuration or tends to be overloaded as causing a virtual network function. Application 200 performance degradation problem bottleneck.

Compared with the conventional technology in which the maintenance personnel of the virtual platform manually collect performance data, the virtual function performance analysis system 100A (or 100B) disclosed in this case collects and analyzes the data of each performance indicator in an automated manner and is timely. and speed advantages. In the conventional technology, the maintenance personnel of the virtual platform do not focus on the virtual platform. The performance index "execution time" of the virtual function function is analyzed, so the cause of the abnormal operation cannot be accurately analyzed; and the virtual function performance analysis system 100A (or 100B) of the present disclosure can accurately analyze the execution time of each virtual function function. . Moreover, compared with the conventional technology which only locates the bottleneck of performance degradation to the virtual network function application, the virtual function performance analysis system 100A (or 100B) disclosed in this case can more accurately locate the bottleneck of the problem to the virtual network. specific functions in the functional application, and can accurately analyze the reasons for the performance degradation of the virtual platform 400.

In addition, in view of the reasons for the decline in virtual platform performance, the virtual function performance analysis system 100A (or 100B) disclosed in this case can also automatically adjust the services of the virtual network function application 200 and/or virtual function functions 202, 204, and 206. Or automatically expand or adjust the configuration of physical resources and/or virtual resources of the virtual platform 400 to automatically improve the performance of the virtual platform 400.

Although the present invention has been disclosed in detail with embodiments and examples as above, it should be understood that these examples are meant to be illustrative rather than restrictive. It is expected that those with ordinary skill in the art can think of various modifications and combinations, and the various modifications and combinations fall within the spirit of the present invention and the scope of the appended patent application.

104:Monitoring unit

200:Virtual Network Function Application

202,204,206:Virtual function function

202a,204a:queuing

T_02, T_04, T_06: Execution time

400:Virtual platform

Claims (16)

A virtual function performance analysis system includes: a monitoring unit for monitoring the performance performance of at least one virtual function of a virtual network function application on a virtual platform, the virtual platform having at least one physical resource and at least one virtual resources, each of the physical resources, each of the virtual resources and each of the virtual function functions has at least one performance indicator, each of the performance indicators has an expected value and/or a threshold, and the monitoring unit is monitored on the virtual platform And record an actual value of each performance indicator; and a performance analysis unit for comparing the actual value of each performance indicator with its corresponding expected value and/or threshold to obtain a comparison result of each performance indicator and based on To analyze system performance, if the comparison result is that the actual value of each performance indicator is higher than the corresponding expected value and/or the threshold, then the performance analysis unit determines the entity resources to which each of the aforementioned performance indicators belongs. , each virtual resource or each virtual function function is a performance bottleneck of the virtual platform. The virtual function performance analysis system as described in claim 1 further includes: a test unit for monitoring each virtual function function of the virtual network function application on a test platform and creating each virtual function function. The expected value and/or the threshold of each performance indicator. The virtual function performance analysis system as described in claim 1 further includes: a resource adjustment unit, if the comparison result is that the actual value of each performance indicator is higher than the corresponding expected value and/or the threshold, then The resource adjustment unit performs at least one of the following based on the comparison result: adjusts the service of the virtual network function application, adjusts the configuration of each physical resource, or adjusts the configuration of each virtual resource of the virtual network function application . The virtual function performance analysis system as described in claim 1, wherein the performance index of each virtual function function is an execution time, an execution frequency, an error rate or a queue length of each virtual function function. At least one of the performance indicators of each physical resource and each virtual resource is a respective usage amount. The virtual function performance analysis system as described in claim 4, wherein if each virtual function function is executed in an environment of multiple execution threads, the execution time, the execution frequency, the error occurrence rate or the queue The length refers to the execution time, execution frequency, error rate or queue length of each virtual function function corresponding to each thread. The virtual function performance analysis system as described in claim 4, wherein if the comparison result is at least one of each of the physical resources and/or each of the virtual resources If the usage is approaching the upper limit of the configuration or may be overloaded, the resource adjustment unit will perform at least one of the following based on the comparison result: expand the configuration of each physical resource that may be overloaded, or adjust the virtual resource Each virtual resource of the network function application is configured so that the virtual network function application and each virtual function function have sufficient resource allocation. The virtual function performance analysis system as described in request 4, wherein, if the comparison result is that the usage of each physical resource and the usage of each virtual resource do not approach the upper limit of the configuration and do not tend to be overloaded , then the performance analysis unit determines that each virtual function function to which each performance indicator belongs has a program logic error or a packet processing exception. The virtual function performance analysis system of claim 1, wherein the physical resources of the virtual platform include at least a physical CPU, a physical memory, a physical network bandwidth, and a physical disk input/ Output, at least one of the above, and the virtual resources of the virtual platform include at least a virtual CPU, a virtual memory, a virtual network bandwidth, and a virtual disk input/output, at least one of the above By. A virtual function performance analysis method includes: monitoring the performance performance of at least one virtual function function of a virtual network function application on a virtual platform, the virtual platform having at least one physical resource and at least one virtual function Virtual resources, each physical resource, each virtual resource and each virtual function function each have at least one performance indicator, each performance indicator each has an expected value and/or a threshold value; monitor and record each on the virtual platform An actual value of the performance indicator; compare the actual value of each performance indicator with its corresponding expected value and/or threshold to obtain a comparison result of each performance indicator and analyze the system performance accordingly; and if the comparison result is Because the actual value of each performance indicator is higher than the corresponding expected value and/or the threshold, it is determined that each physical resource, each virtual resource or each virtual function function to which each aforementioned performance indicator belongs is the Performance bottlenecks of virtual platforms. The virtual function performance analysis method described in claim 9 further includes: monitoring each virtual function function of the virtual network function application on a test platform; and creating each virtual function function on the test platform. The expected value and/or the threshold of the performance indicator. The virtual function performance analysis method as described in claim 9, wherein if the comparison result is that the actual value of each performance index is higher than the corresponding expected value and/or the threshold, then the virtual function performance analysis method Also includes: Adjust the service of the virtual network function application, adjust the configuration of each physical resource, or adjust the configuration of each virtual resource of the virtual network function application based on the comparison result. The virtual function performance analysis method as described in claim 9, wherein the performance index of each virtual function function is an execution time, an execution frequency, an error rate or a queue length of each virtual function function. At least one of the performance indicators of each physical resource and each virtual resource is a respective usage amount. The virtual function performance analysis method as described in claim 12, wherein if each virtual function function is executed in an environment of multiple execution threads, the execution time, the execution frequency, the error occurrence rate or the queue The length refers to the execution time, execution frequency, error rate or queue length of each virtual function function corresponding to each thread. The virtual function performance analysis method as described in claim 12, wherein if the comparison result is that the usage of at least one of the physical resources and/or the virtual resources is approaching the upper limit of the configuration or may be tends to be overloaded, the virtual function performance analysis method further includes: Expand the configuration of each physical resource that may be overloaded based on the comparison result, or adjust the configuration of each virtual resource of the virtual network function application so that the virtual network function application and each virtual function function have sufficient The amount of resource allocation. The virtual function performance analysis method as described in claim 12, wherein if the comparison result is that the usage of each physical resource and the usage of each virtual resource do not approach the upper limit of the configuration amount and do not tend to be overloaded , the virtual function performance analysis method further includes: determining whether each virtual function function to which each performance indicator belongs has a program logic error or a packet processing exception. The virtual function performance analysis method as described in claim 9, wherein the physical resources of the virtual platform include at least a physical CPU, a physical memory, a physical network bandwidth, and a physical disk input/ Output, at least one of the above, and the virtual resources of the virtual platform include at least a virtual CPU, a virtual memory, a virtual network bandwidth, and a virtual disk input/output, at least one of the above By.

Images